@spec gen_ai_latency_e2_e(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_model_decode(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_model_execute(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_model_forward(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_model_inference(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_model_prefill(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_queue(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_in_scheduler(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_latency_time_to_first_token(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_request_id(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_request_max_tokens(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_request_n(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_request_temperature(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_request_top_p(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_response_model(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_usage_completion_tokens(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_usage_num_sequences(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec gen_ai_usage_prompt_tokens(SnakeBridge.Ref.t()) ::
  {:ok, term()} | {:error, Snakepit.Error.t()}

@spec new(
  [term()],
  keyword()
) :: {:ok, SnakeBridge.Ref.t()} | {:error, Snakepit.Error.t()}